All models are wrong - more or less! Added value of model calibration for dry and wet weather in Munich

Hydraulic models form the basis for far-reaching investment and operational decisions as well as emission- or immission-based considerations. Model calibration and validation are essential to ensure adequate model quality and to assess model uncertainties, particularly in large and complex networks. This study presents the methodology for the calibration of Munich's extensive hydrodynamic sewer network model, comprising 2,400 km of sewers and serving an area of 18,000 hectares. The calibration process is based on a comprehensive 18-month measurement campaign, which included flow, water level, and rainfall data collected from more than 500 temporary and permanent monitoring stations. The calibration process identified and corrected several model inaccuracies, including errors in flow paths, connection elevations, operational settings and assumptions on infiltration water discharge. The results demonstrate the critical role of measurement data in refining hydraulic models for large, highly meshed networks. In a future workflow, quality measurements will be included in the calibration process. The project impressively shows that (all) models are more or less wrong if not calibrated to measurement data. How wrong can only be answered within the framework of a model calibration by proofed measured data and thus whether the model is suitable for the respective application.